AI Agents for Sustainable Livestock Farming

A Firecrawl growth team member, Cody Menefee, prompted his audience to look past conventional software development toward larger-scale problems during his opening remarks. He introduced his talk by declaring, "Today I'm here to tell you you're not thinking big enough."

Menefee emphasized the need to tackle problems that extend beyond building software for other software developers, highlighting a common pitfall in the tech industry. He lightheartedly acknowledged a title overlap with a video by Theo, suggesting a shared sentiment about expanding the scope of innovation.

He clarified that the discussion would center on actual farming, particularly livestock, and the automation of managing animals on pasture.

While admitting his personal background is in manual labor and software development rather than professional farming, Menefee underlined the critical role of technology in this often-overlooked sector. The core challenge involves optimizing the movement of animals to maintain pasture health and animal welfare.

Menefee introduced himself as a 'vibe coder,' emphasizing his reliance on AI agents for development rather than memorized syntax. He stated, "I use AI agents all day long. I don't have any syntax memorized. I am not proficient in any particular coding language, but I will crank out some stuff on a weekend."

Currently at Firecrawl, Menefee's work involves building web data APIs that provide essential context for AI agents. This approach allows individuals without deep traditional coding proficiency to create robust solutions by leveraging AI's capabilities.

Menefee shared a personal anecdote about transporting ten turkeys in his Tesla to a processor, illustrating the hands-on reality of small-scale farming. His experience highlighted the significant economic hurdles, as he realized the difficulty of making farming profitable.

He explained his decision to pivot from full-time farming to applying engineering principles to scale agricultural systems. Menefee noted the personal sacrifices involved in traditional farming, stating, "I realized it's actually really hard to make any money farming. Surprise, surprise. And I have a wife. I have two kids. It didn't feel right to ask them to give up the lives they had so that I could go cosplay as a farmer and raise chickens."

This shift stemmed from a desire to leverage technology to create a more viable and scalable model for sustainable agriculture. He found more leverage in applying his technical skills to these systemic issues rather than continuing with labor-intensive, small-profit operations.

Menefee critiqued the software industry's insular focus, where many developers build tools primarily for other developers. He urged the audience to seek out and solve problems in tangible, foundational sectors rather than focusing solely on more Software-as-a-Service (SaaS) solutions.

He challenged the prevalent trend, saying, "I really am just here to challenge you that there are other problems to solve than just another MCP for another SaaS solution at another company." The ultimate goal is to redirect technical efforts toward impactful, real-world challenges, such as those found in agriculture.

Menefee passionately advocated for livestock to be raised on pasture, asserting that it benefits animals, consumers, farmers, and the ecosystem. He stated, "I believe livestock belongs on pasture. I think animals should live on grass. Um I think it's better for the animal, the consumer, the farmer, the ecosystem."

Currently, approximately 97% of cows in the U.S. are finished in feedlots, with only 3% raised on pasture. The presentation's primary objective is to significantly increase the percentage of animals that are raised on grass, shifting away from industrial feedlot practices for more sustainable and humane animal husbandry.

He explained that raising animals on grass, particularly with proper rotational grazing, is labor-intensive, requiring constant movement of animals.

Simple grazing on a single large plot degrades pasture quality due to selective grazing and trampling. Rotational grazing, which involves moving animals daily between smaller paddocks to allow for rest and regrowth, addresses this issue but demands significant manual effort to manage effectively.

Companies like Halter, valued at $2 billion, and No Fence are developing GPS-connected collars for virtual fencing. These technologies enable farmers to draw virtual boundaries and remotely move animals, significantly reducing physical labor associated with traditional fencing.

However, these tools do not solve the complex decision-making problem of determining the optimal location to move animals. Current grazing decisions rely heavily on farmers' intuitive assessments of environmental conditions such as drought, rainfall, and plant growth cycles, which remain a key challenge for automation.

Farmers must carefully manage grass height to prevent either stunting growth through over-grazing or allowing it to become bitter and unpalatable due to over-maturity. The goal is to keep grass in its 'juvenile sweet spot' to maximize its nutritional value and promote regrowth.

Achieving this balance requires continuous movement of animals across pastures. This ensures that the grass is consistently grazed at its most optimal growing age, which is crucial for maintaining pasture health and maximizing overall productivity.

Drone orthomosaic maps are proposed as a method to monitor grass health and quality through high-resolution imagery. Automating drone flights could allow for extensive data collection across pastures, providing valuable insights for management decisions.

However, implementing this technology faces significant challenges, including the need for farmers to acquire new skills, unresolved regulatory issues, and a lack of approved autonomous drone flights worldwide. These technical and governmental bottlenecks make drones a less feasible near-term solution compared to existing satellite alternatives.

While companies like Planet provide daily 1x1 meter resolution imagery of the entire globe, the high altitude of satellites limits the specificity required for precise farm-level management decisions. It is challenging to discern detailed conditions necessary for effective grazing strategies.

Field-level researchers at Missouri Lincoln University use trail cameras to measure grass height with precision, highlighting the need for ground-level intelligence. Such reference points are essential for accurately tracking vegetation regrowth and making informed decisions on pasture management.

This system would allow an LLM to make optimal decisions on animal placement daily.

The LLM would process inputs such as GPS location, historical movement data, drought conditions, and total grass height. By calculating optimal positions for individual cows, the entire herd, and the broader ecosystem, this system aims to increase stocking density on fewer acres, making pasture-based farming competitive with industrial feedlots.

Menefee identified three main blockers preventing the creation of a fully autonomous grazing system: building a comprehensive knowledge base, developing a visualization layer, and addressing proprietary geofencing technologies. He stated, "There's three main blockers that I think need to exist in order for us to actually create this system."

The knowledge base involves scraping YouTube agricultural channels and research papers, while the visualization layer is needed to assess biomass and pasture biodiversity. Furthermore, the issue of proprietary geofencing collars, which lock users into closed ecosystems, must be resolved. Tools like Firecrawl and Open Pasture are designed to centralize and share this critical farming knowledge to overcome these challenges.

Menefee criticized current geofence providers like NoFence and Halter for their proprietary hardware-software models, which he believes hinder industry-wide innovation. He acknowledged the value of their technology but strongly disagreed with their closed approach.

He explained, "in order to use their software they require you buy their collars, and you can't plug your own software into their collars." From an industry standpoint, Menefee argued this coupling creates significant friction. His goal is to promote an open ecosystem where third-party software can integrate with existing tracking collars, fostering greater competition and optimization in livestock management.

He reiterated his criticism of proprietary software and emphasized the need for repairable hardware in the agricultural sector. An open-patent and open-API hardware design is crucial to foster competition, allow for continuous optimization, and prevent vendor lock-in, enabling broader adoption and innovation in autonomous grazing.

He explained that when ruminants like cows, sheep, or goats graze first, they trim the top of the grass, preparing the pasture for subsequent species.

Chickens then follow, consuming parasites from the droppings, which creates a synergistic relationship that reduces medication costs and improves overall herd health. This practice, exemplified by systems like Pasture Bird's automated coop, which moves every 24 hours, enhances pasture health and productivity while minimizing external inputs.

He proposed deploying LLMs as reasoning agents within multi-data input systems to assist farmers with decision-making.

By allowing LLMs to process diverse data and suggest actions that humans can confirm or deny, the goal is to overcome the limitations of farmers' personal labor and decision-making capacity. This approach aims to scale operations, enabling more animals to be raised on grass and promoting sustainable agricultural practices.

He stated, "The number one problem is actually just giving the agents the context they need, gathering the data, packaging that data, and then presenting it in a way that the LLM can reason over."

Firecrawl's mission is to build this 'context layer' for AI by gathering and structuring web data, making it accessible for LLMs to process. Menefee concluded with a call to action for engineers to contribute to solving these complex real-world problems and invited collaboration through Firecrawl or Open Pastures.